Efficient object detection speeds up fire smoke monitoring on devices
YOLOF-MLD: A Multi-Level Distillation Framework for Efficient Object Detection
Computer Vision and Pattern Recognition
Summary
Detecting objects quickly and accurately on small devices like Raspberry Pi is important for things like spotting fire or smoke. The authors show how replacing a complex part of a detection system with a simpler one can make it run faster but usually hurts accuracy. They fixed this by using a clever training method that teaches the simpler system to mimic the detailed one at multiple levels. Their method runs much faster on a Raspberry Pi with almost the same accuracy, making it better for real-world safety devices.
What this means in practice
- •For embedded system engineers: Run fast and accurate fire and smoke detection on low-power hardware by using a distilled YOLO framework designed for single-input outputs.
- •For smart home device makers: Create efficient safety monitoring products that detect hazardous conditions quickly on edge devices with limited computing resources.$Commercial implications: Enables safer, faster home safety products with advanced fire and smoke detection powered by lightweight neural networks.
Authors
Xiaolong Wu, Liming Zheng, Zitong Liu, Yue Zheng
Abstract
Efficient object detection on resource-constrained edge devices is essential for safety-critical applications such as fire-smoke monitoring. However, current lightweight frameworks, such as YOLO26, still rely on the parameter-dense multi-branch Feature Pyramid Network (FPN), preventing full exploitation of their lightweight advantage. Replacing FPN with YOLOF's single-input-single-output (SiSo) neck is a promising solution but causes substantial accuracy loss. To alleviate this, we propose YOLOF-MLD, a CPU-efficient multi-level distillation (MLD) framework that combines a YOLO26n backbone, a SiSo neck, and MLD. It aligns features at the backbone via sliced-semantics matching, neck with receptive-field matching, and response levels. On D-Fire and Annotated Fire-Smoke (AFS) datasets, YOLOF-MLD reduces the parameter count of YOLO26n by 35.7\% and achieves a 62.3\% speedup on Raspberry Pi 5, attaining 71.6% and 59.9% mAP@50, respectively, compared with 75.5% and 60.0% for YOLO26n. These results demonstrate a competitive accuracy--latency trade-off for edge deployment. Our code is available at https://github.com/XiaolongWU111/YOLOF---MLD/blob/main/README.md.